Statistically Validated Indices for COVID-19 Public Health Policies

In this paper we present six new indices generated from a Bayesian measurement model that allow us to combine policy data from two of the most comprehensive COVID-19 policy datasets, the CoronaNet COVID-19 Government Response Event Dataset and the Oxford COVID-19 Government Response Tracker. By doing so we can validate both independent sources of data and provide more information than either dataset on its own. We estimate these time-varying indices with summary scores for each day from January 1st, 2020 to January 15th, 2021 for over 180 countries and six policy categories: social distancing policies, school-related policies, business-related policies, health monitoring policies, health resources policies and mask-related policies. We also estimate models that predict these indices with a range of social, public health, political and economic covariates. Our results show that business restrictions and social distancing restrictions are strongly associated with reduced general anxiety while school restrictions much less so. Furthermore, school restrictions are associated with higher rates of personal contact with people outside the home, higher levels of income inequality and bureaucratic corruption. Finally, we also find that heads of state who are women are more likely to implement a broad array of pandemic-related restrictions than male leaders.